Should-cost versus your real cost price

What is a should-cost model?

A should-cost model is a calculation of what a part ought to cost, made by your customer's purchasing or engineering team. It adds up material, machining time, setup, tooling, overhead and a profit allowance. Some customers have a cost office behind it, with software that turns a 3D model and a drawing into a price.

UK government guidance describes it as a forecast of what something should cost, with market factors such as risk and profit included. It distinguishes top-down, from analogous projects or a parametric model, and bottom-up, by adding up costs at the lowest level.

In a 2024 master's thesis at the University of Twente, an interviewee from an organisation with its own cost office describes the same split: a macro or parametric should-cost based on a few characteristics of the part, and a detailed or analytical should-cost per component or subassembly. Organisations in that study use should-cost to test a supplier's price and to negotiate it.

A should-cost is not a target price. A target price is a number; a should-cost model is a build-up you can open up and argue with.

Where do the buyer's numbers come from?

A should-cost model is built without your shop. Typically it derives a route from the geometry, picks a standard machine, estimates cycle time from the volume to be removed and multiplies it by a rate from a database. On top come an assumed batch size, a percentage for overhead and a profit allowance.

The rate alone differs by country. Eurostat puts average hourly labour costs across the whole economy in 2025 between €12.0 in Bulgaria and €56.8 in Luxembourg, with €47.9 in the Netherlands and €34.9 for the EU as a whole. A machine hour rate is more than wages, but a model priced on a European average, or on another country's rate, will price the same hour differently from your books.

So ask for the assumptions first. The UK guidance comes with a template for exactly this, the book of assumptions: a log of every assumption and data point in the model. A buyer working from a model can tell you which machine, cycle time, quantity and rate it was run on.

Where does the model part ways with your shop?

The gap rarely sits in one big number. It sits in a handful of lines where the model assumes an average and you hold a fact. These are the lines to compare first.

• Material: the model often prices the net volume or a thin allowance. You buy stock in a standard size. From it you saw a blank with room to clamp.
• Setups: the model often assumes a machine that makes the part in one setup. On your machines it may take two, each with its own setup time.
• Cycle time: a different machine, a different tool or a thin wall that has to run slower all change the time per part.
• Rate: the model uses a rate from a database; you use the hourly rate of the machine that actually does the work.
• Batch size: setup and programming time spread over the quantity. If the model assumes a larger batch than the order, that one-off share per part comes out too low, as a small batch per part shows.
• Finishing: an anodiser often charges a minimum per batch. The model then carries only a price per part, with no subcontract line holding that minimum.
• Inspection: if the drawing asks for a first article inspection or a measurement report, check that the model carries the inspection time.

Worked example: one housing, two cost prices

An example in round numbers: a milled housing in EN AW-6082, 50 parts, black anodised. The should-cost model assumes one setup on a 5-axis machine at €80 an hour. The shop makes the part on a 3-axis machine in two setups, at €90 an hour. Both columns show cost price, before profit.

The gap of €13.50 a part is not in one line. Material, setup, programming and anodising account for €8.50 of it, more than the €5 in machining. At 200 parts the gap narrows to about €9: setup and programming spread over more parts, and the anodiser's minimum no longer applies.

What is left is the real conversation: €3 of material from a standard stock size and €5 of machining, two minutes longer on a machine with a higher rate. Those are facts about your shop, not margin.

How do you answer a should-cost line by line?

Total against total ends in an argument about percentages. Estimate line against estimate line ends in a conversation about facts. This is how to run it.

1. Ask for the model and its assumptions: route, machine, cycle time, quantity, rate, and the allowance for overhead and profit.
2. Set the model's lines beside your own, in the same order: material, machining, setup, programming, finishing.
3. Mark for each line whether the gap is in time, rate, quantity or a supplier's price.
4. Show the inputs on every line of yours: time, rate and quantity, and the source of every purchase or subcontract price.
5. Offer what genuinely closes the gap: a quantity break at a larger batch, a different setup, or one combined batch for finishing.
6. Keep cost price and margin apart. Whatever you take off the price after that is a pricing decision, not a lower cost price.

Which lines do you concede and which do you defend?

Sometimes the model is right. If it shows a route in one setup where you need two, that is information about your own work: perhaps the part belongs on another machine, or a dedicated fixture pays for itself on repeat work. A cost engineer who concedes that point earns credibility for the lines that stay.

Defend the lines that come from your own shop: the hourly rate per machine, the material you actually buy and your anodiser's minimum. Do not go below your cost price to meet the model. Buyers with a model distrust a price that is too low as well: under the UK guidance, on a complex project a department must refer any bid more than 10 per cent below either the should-cost estimate or the average of the other bids to a central team before accepting it.

A risk allowance belongs on the table too, as a line of its own rather than buried in the rate, with its reason attached: a tight tolerance, a thin wall, a first batch. A risk allowance with no reason is the first thing a buyer strikes out.

Who has the best model of your shop?

Should-cost software is built for the engineering and purchasing departments of customers. Quoting platforms show a price straight after an upload, then look for a shop that will make the part at that price. Both price an average machine shop, and the average machine shop does not exist.

The supplier holds what neither of them has: the machines on the floor, the rates from its own books, the setups its manufacturing engineer chooses and the reasons behind each choice. Put that knowledge into one model of the shop and a should-cost stops being a threat. You answer it with an estimate the buyer can check.

Blake builds on that. The next step is calibration on actual hours, so that every time in the estimate demonstrably comes from your own production. The step after that puts the same lines in front of the designer before the drawing is frozen. As a supplier you will then argue setups and tolerances while they can still change, rather than percentages after the fact.